The boundary layer on modern profiles with flaps
by Karel Termaat
Introduction
Have you ever heard of a boundary layer of wing
profiles and is this of interest to you as a glider pilot. Yes, I think so. With its typical laminar and
turbulent forms and characteristic transition process and consequences on drag,
I think this is an interesting aerodynamic phenomenum to learn more about. The
existance of a boundary layer forms the basis for the so called “laminar
bucket” in the drag curves of wing profiles. Operating the glider properly
within the limits of this low drag bucket is essential in obtaining optimum L/D
performances of the wing during climb as well as during straight glides.
Proper mathematical formulations to describe all
details of boundary layers, either in its laminar or turbulent form have been
developed just recently. The transition
process from laminar to turbulent is now well understood. In the cause of time,
detailed experimental results in modern wind tunnels and during actual flights
have been obtained and support the theoretical models.
Developments
- From the past to now
Even before the beginning of our era, Aristotle and Archimedes studied the forces acting on floating and submerged
objects in still water. From the middle
ages on, also da Vinci, Gallileo, Newton and Bernoulli did that also,
but especially in running water. From their reflections and experimental work
with the primitive means which they had, they noted already in the 17 century
that flow around a submerged object
causes a total force on the object that can be broken down in a component
perpendicular to the direction of the flow (the lift) and a component in the
direction of the flow itself (the drag). A for that early time notable and
correct observation. Not much, also first mathematical relationships for the
size of those forces were developed.
Nowadays both forces are described with the
well-known formula for lift L = ½ ρ V 2. S.CL and drag D = ½ ρ V 2. S.CD in a modern way. The two formulas, which in fact contain all aspects of
the stable hydrodynamic or aerodynamic forces of lift and drag are absolutely
simple. However the difficulty lies in the description of the lift coefficient CL and the drag coefficient CD in these
formulas. The coefficients are controlled by the shape of the object and the
angle at which the water or air flows onto this shape. They can only be
calculated directly with very detailed knowledge of the local distributions of
static pressure and friction and the availability of modern theory and
computers. Of course it was impossible in the 17 century to measure those
distributions, but even today advanced software and computing power is required
to find the coefficients of lift and drag from these measurements if available.
However if in a test-rig, you measure the size of the lift L and
the drag D at several angles of attack accurately, than you can
with V, S and ρ known, calculate
CL and CD from the rewritten versions of the above formulas, where CL = L/( ½ρV 2 S) and CD = D/(½ ρV2S). The results for CL and CD can
comfortably be presented in charts as a
function of the angle of attack Alfa. In the past, it was exactly done in this way
using simple experimental set-ups with e.g. thin wooden boards in running
water. But also during these times detailed measurements on 2D wing sections
are done with the same goal in modern wind tunnels like the one at TUDelft. An
early example of lift curves obtained in this way is given in Figure 1,
while Figure 2b shows lift curves for a modern profile.
Figure 1: Lift curves CL - Alfa (ref. J.D. Anderson, Jr.)
Drag curves are usually not presented as a function of Alfa. Instead, pairs of CL - CD
values are plotted on a graphic sheet sheet for each possible flap position
together with the associated lift curves. With these presentations one obtains
per flap position, which in fact means ever selecting new profile forms, a good
overview of the low drag area of the profile, the so called “laminar bucket”,
together with the relevant lift curves. Figure 2a and 2b give those combinations for one of the wing
profiles of the modern Antares.
Figure 2a and 2b: Measured C L - CD
curves and C L -Alfa curves of one of
the profiles
of the wing of the Antares for five flap positions at Re = 1.5 E6 (ref. LB, TUDelft)
- Theoretical progress
- Principia Mathematica:
In 1687, the famous mathematician and physicist Isaac
Newton defined in his publication "Principia Mathematica"
three important laws:
1. an object in motion remains in that same motion
unless a force works on it: V2 = V1
2. an object with mass m is accelerated by
a force F proportional to the size of that force: a = F/m,
better known as F = m.a
3. if an object A acts with a force F
on object B than that happens also vice versa: F B =
-FA
On the basis of these "Principia" and
the basic knowledge of the later formulated laws of conservation for mass,
momentum and energy, it would already at that time have been possible to
describe the behavior of packages of flowing water or air mathematically. But
Newton was unsuccesfull in doing so and in fact unintentionally delayed by his
somewhat dominant authority the further unfolding of fluid mechanics for a
while.
- Inviscid flow:
Only after the beginning of the development of
modern mathematics, Euler came for inviscid (non viscous) flow in
1755 with a correct partial differential equation for the calculation of flow
velocities and pressure fields on the basis of Newton's second law. Shortly
therafter however d'Alembert proved that with those equations objects will of course have no
frictional drag, but also no pressure drag (paradox of d'Alembert). Apparently
the viscosity of the fluid may not remain unattended.
- Viscous flow:
Around 1830 Navier and Stokes created at about the same time
an additional term in the Euler equation such that the viscosity and hence
friction and pressure forces are taken into proper account. With this an Euler
system of partial differential equations becomes even more complicated of
course. Today, with fast computers and much computation time the equations of
Navier-Stokes can be solved in an iterative process using small coupled volumes
covering the entire fluid domain. Appropriate software , such as FLUENT, is
currently still quite expensive and requires a lot of computation time. Thanks
to powerful computers, this software is increasingly applied at the larger
institutes. Until recently a dilemma in Navier-Stokes calculations was that in
the domain considered the flow may not be partly laminar and partly turbulent,
and that is typically precisely the case with flows close to the surface of
wing profiles. But the developments go fast and N & S is certainly the tool
of the future in the aerodynamics of gliders.
- A grandiose idea:
In 1904, Prandtl came upon the
excellent idea to devide the flow domain into a large outer region with
inviscid flow and a thin boundary layer attached to the surface of a profile in
which the flow is viscous. Through adhesion this boundary layer sticks to the
surface, while just a little above the surface the velocity in the boundary
layer adjusts to that of the local outer flow. In the outer region velocity and
pressure distributions can be calculated using the some what less complicated
equations of Euler for inviscid flow, while in the thin boundary layer with
viscous flow those distributions and shear stresses working on the surface can
be obtained using simplified forms of the Navier-Stokes equations. On the basis
of this idea Prandtl and his students such as Blasius, von
Karmann and Schlichting at the University of Göttingen
soon came up with interesting results. In fact strongly needed for the further
development of aircraft design and construction after the work of the first
pioneers like Otto Lilienthal and Wilbur and Orville
Wright. Still today, this dual approach is the best way to study in
detail properties and effects of flows around wing profiles at low velocities.
At the higher velocities shock waves occur in the flow pattern and the unique
Prandtl approach can than not be applied.
- The flow pattern around a wing profile
For the calculation of the flow pattern around a
profile with inviscid flow the fluid domain is described with a large number of
coupled two-dimensional volumes of variable size and negligible thickness. In
the direction perpendicular to these flat volumes, there is no flow. Close to
the profile and especially where the contour of the profile strongly changes,
such as at the front and the rear of the profile, a fine-grained grid is
necessary to obtain accurate results when calculating flow velocities and
pressures. Figure 3 gives a simple example of such a 2-D grid.
Figure 3: 2-D volumes around a profile
A boundary layer is not supposed yet. The entire
domain has inviscid flow, therefore in all the 2-D volumes the differential
Euler equation can be applied. The total set of these differential equations is
solved iteratively together with the equations of conservation of mass and
momentum that describe the coupling between the small volumes. After several
iterations the calculated velocity and pressure field converge to a closed
final solution and the effect of the shape of the profile on the velocity
pattern becomes clearly visible. Corresponding values of velocity and pressure
can be indicated on streamlines around the profile. This has been done in
symbolic form for one position (x, y) in figure 4. This figure also shows
several other aspects of a calculated flow pattern around a profile, however
the effects of a thin boundary layer is not considered.
Figure 4: Stream line pattern around an airfoil with inviscid flow
- The potential theory
Even before the above theory of Euler could be
applied in practice, potential theory for inviscid, rotation free flow was
developed in analogy with definitions
used in electro-technique. A surface of for example a wing profile, is covered
with a dense grid of small flat or curved panels, where in control points of the panels small flow
sources and eddies are supposed to exist. Those sources and eddies affect the
flow pattern in each and every position of the domain, particularly in the
control points of all the panels considered. In an iterative process of matrix
calculations with boundary conditions, the velocity pattern that is obtained in
this way converges quickly to a model that fits exactly to the shape of the
profile.. Experience has shown that the results of this type of calculations
fit stunningly well to practical flow patterns found on wing profiles used in
e.g. the sport of soaring. Where only velocitys and pressures on the surface of
profiles are of importance, the potential theory is therefore frequently applied rather than the theory of
Euler. But in special situations results of the potential theory are not
accurate enough and Euler must still be applied.
Kutta and Joukowski had
noticed in a global consideration earlier on that all eddies supposed in the
control points of the mini panals used in potential theory cause a circular
flow around the profile, the so called “circulation”. When added to the main
flow, this circulation determines the lifting capacity of the profile, bearing
in mind that the law of Bernoulli (Pdyn + Pstat = constant)
translates unequal flow velocities into pressure differences between the upper
side of the wing and the lower side.
- Designing wing profiles
Nowadays fast computational methods for the design
of wing profiles are available. Well-known 2D codes based on potential theory
are XFOIL ( Mark Drela, MIT 's) and PROFIL ( Richard Eppler's,
Stuttgart). Both codes and extensions thereof in the 3D domain such as the
VSAERO code, are used to thoroughly analyze existing profiles or to design new
profiles by careful geometric adjustments of a reference profile to achieve
intended aerodynamic properties The velocity vector fields that these codes
compute just on the outside of the boundary layer, is required to find pressure
distributions over the surface of the profile through the law of Bernoulli.
These velocity fields are also required as a boundary condition to solve the
equations that apply to the thin viscous layer between the profile surface and
the mainstream, as introduced by Prandtl in 1904. These boundary layer
solutions influence by their so-called
displacement thickness the calculated velocity field in the inviscid flow
region, so here a coupled iterative computational process exists.
Usually the computed pressure distributions are
graphically presented in the form of the pressure coëfficient Cp
as a function of profile depth x/L. In figure 5 an
example of such a distribution for one of the profiles of the Concordia wing is
given. Some typical flow data are listed too. It is striking that the flow rate
on the profile is soon about 50% higher than that of the undisturbed flow in
front of the wing, while the local pressure at that point is only about 1%
lower. For an aerodynamicist, these graphical presentations are very
informative when searching for a profile that fulfills aerodynamic
expectations. In fact one can find these presentations for all studies on wing
profiles. Figure 8 and also 9 are typical examples of these
pressure profiles.
Figure 5: Calculated and measured pressure curves of one of
the Concordia profile with flaps = 0 (Ref. LB, TUDelft)
-Wind tunnel measurements
By repeating the computer calculations for
different angles of attack and flap positions one can see the effects on the
pressure curves quite well. As said the local pressures on the profile are
computed by the programs from the local velocities through the relation of
Bernoulli. These local pressures create force vectors perpendicular to the surface
of the profile which can be split up into components perpendicular to the
incoming flow and components parallel to that flow. By integration of the
perpendicular components over the perimeter of the profile one obtains the
total lift while the total drag
experienced by the profile follows from integration of the pressure components
along the shape of the profile.
The theoretical lift coefficient CL and drag coefficient CD are obtained
directly from these calculations as a function of the angle of attack α
for all flap positions considered. Usually accurate wind tunnel measurements
are carried out for selected profiles to verify calculated pressure curves and
lift- and dragcurves experimentally. Figure 2a and 2b and the
dots in figure 5 are examples. Typical additional difficulties
encountered in the practical application of profiles are the complicated flow
patterns at the wing root and that associated with the free end of the wing.
The application of aerodynamic well-designed fairings and the
use of optimised winglets therefore receives much attention nowadays.
Boundary layer calculations:
- Velocity distribution in the laminar boundary
layer
As said, Prandtl defined a thin boundary layer with
viscous flow for which simplified forms of the Navier-Stokes equations can be
obtained by deleting less important terms. These boundary layer equations,
later named the Prandtl equations, can in their differential form nowadays
easely be solved by the computer. Boundary conditions are that at the surface
itself the flow velocity is zero (no slip) and that at the top of the boundary
layer the velocity is for 99% is equal to that of the local flow just outside
the boundary layer. For a flat plate the velocity and pressure in the direction
of the outer flow is constant and than the Prandtl equations simplify even
further. From this Blasius came up with a true analytical formulation for the
velocity distribution perpendicular to the surface. Figure 6 is an image
of that. Initially the velocity increases linearly with the distance from the
surface, while when approaching the full thicknes of the thin boundary layer
the velocity adopts to that of the free
flow just outside the layer. In the direction of the flow this perpendicular
velocity distribution has a constant shape while the thickness of the boundary
layer increases from very thin until a few millimeter as long as the flow is
laminar. At a certain distance from the beginning of the plate the flow in the
boundary layer becomes unstable. Small wavy eddies appear which increase gradually in size while flowing
downstream. At a certain location they explode as it were, where the flow
transitions from the laminar form to turbulent. The simple perpendicular velocity distribution than loses its characteristic
laminar form.
Figure 6: Perpendicular velocity
distribution according to
Blasisus in the laminar boundary layer of a flat plate
With a wing profile, the perpendicular velocity distribution
in the boundary layer changes downstream somewhat its shape because of local velocity and pressure changes of the
main flow. Several researchers like Pohlhausen and Thwaites,
have intensively studied this process. Generally one can say that when the
velocity of the main flow remains constant downstream, the perpendicular
velocity distribution in the laminar boundary layer close to the surface will
evolve as a straight line, as indicated in Figure 6. When the local velocity of
the main flow increases, meaning that the pressure drops, than the
perpendicular distribution at the surface gets a more convex shape and the
boundary layer becomes more stable. When the local velocity decreases causing
the pressure to rise, than the distribution gets a more concave shape and the boundary layer becomes more unstable.
When interpreting calculated or measured velocity distributions of the main
flow one can easily indicate where the boundary layer becomes more stable or just
more unstable. In fact the exact shape of the pressure distribution just
outside of the boundary layer is decisive for the shape of the perpendicular
velocity distribution in the boundary layer and thus decisive for the stability
and detachment effects of that layer. It will be clear that when designing the
geometrical shape of his profile, an aerodynamicist can significantly affect
the place of the detachment of the boundary layer from the profile surface.
- Transition from laminar to turbulent
Allready
Newton figured out that the shear stress resulting from the "sticky"
flow along the surface is proportional to the gradient du/dy of the
perpendicular velocity distribution at the surface of the of the profile as
shown in Figure 6. Directly downstream of the nose of the profile, where the flow
reaches the wing surface for the first time and the boundary layer is still
very thin, the local gradient of the velocity profile is large and much
friction will occur. Further downstream more and more air is slown down and the
velocity distribution in the boundary layer developes as that with a flat
plate. When the velocity in the outside free flow starts slowing down after
having reached a maximum value, the pressure will rise again and at the surface
the gradient of the boundary layer velocity profile may decrease to zero. Than
detachment of the thin boundary layer from the profile surface will occur. The
flow regime in the de-attached boundary layer becomes unstable and changes from
laminar to turbulent. Immediately after this transition the air in the now
turbulent shear layer mixes with the outside flow and is accelerated again. The
new turbulent boundary layer holds stable against the surface of the profile
again. In this transitional process a "laminar release bubble" often
occurs in which one or more small vortices may be present, as tests with
fluorescent oil and special photographic means clearly shows. This bubble may
occur over the entire span of the wing, causing pressure drag and deteriorating
starting conditions of the turbulent boundary layer further downstram. It is
therefore important that in the transition proces the air bubble is kept as
small as possible or even be avoided completely.
In figure 7 the transition process, which
takes place in a viscous layer of air of only a few mm in thickness and is just
a few cm’s in length, is displayed in detail. Some velocity distributions
between the wall and the outside flow as measured with very small pitot tubes
or otherwise are shown. These flow profiles are essential when calculating
aerodynamic properties such as the displacement thickness of the boundary layer
as referred to before. At B one can clearly observe the point where the laminar
boundary layer detaches from the surface and where just downstream also
slightly above the profile, the surface flow comes to a standstill. In the
lower part at C a more or less dead area is present. At D a small clockwise
rotating vortice causes the flow to move backwards over the surface of the
profile. At E the transition of the boundary layer flow to the turbulent form
is largely completed. The figure also indicates that after transition, the
boundary layer with her now turbulent velocity distribution such as at F,
reattaches back again to the surface and significantly increases in thickness.
Figure 7: Transition from
laminar to turbulent with a small laminar air bubble
- Profile depth at the point of transition
A simple, but not very accurate method to find out
when a laminar boundary layer transits to a turbulent form is based on the Reynolds
number. In the not entirely turbulent free air of wind tunnels where on a flat
plate the pressure on the surface equals ambient pressure everywhere,
transition was measured at Re = 2.8 x 106. In perfectly still air of
the free nature this will occur at an even slightly higher Re-value. From these
rather large Re-numbers one can conclude that typically in the velocity regime
of gliders, especially at the lower velocities, the transition of the boundary
layer to the turbulent form takes place late on the profile. The high Reynolds
numbers show in fact that because of the long laminar flow trjectories at most
glider wings they have quite low drag, which is the main reason for the usual
excellent performance of these type of aircraft.
Several researchers have developed analytical
models to calculate how laminar flow detachment from the profile surface can be
delayed as long as possible. They came up
with specific shapes of the profile to avoid early transition of the flow
in the boundary layer to the turbulent form. Best known is the criterion of
Wortman which is a mathematical condition for the pressure gradient along the
chord of the profile. As long as this condition is met, the laminar boundary
layer will stick to the surface. From this condition it follows that the
pressure on the profile, after reaching the minimum may only increase very
slowly as a function of x/L, as
the pressure coefficient Cp in figure 8 shows.
Figure 8: Postponement of transition by a small pressure gradient
Today aerodynamici state that transition will take
place when very small natural variations in the flow pattern, the so-called
TS-waves, have increased by a factor of e 9 (= 8100) in amplitude.
The form of the profile is now tuned downstream in such a way that the
amplification factor in the instability region is as long as possible smaller
than this critical value so that the flow remains laminar. Jan van Ingen
of TUDelft, has since 1957 developed this so-called é-to-the-n-th method.
He isnow very well-known for that. Today his method is generally accepted. Loek
Boermans, also of TUDelft, has frequently applied the e n
method of his teacher in detailed designs of well performing wing profiles with
low drag for gliders and motor planes.
In conclusion, one can say that at some point
after the thickest part, the contour on the upper side of the profile has to
join again that of the lower side. In this part the pressure increases again,
however by applying the criteria of Wortmann and van Ingen this can be done in such a way that the
strength of the TS-waves in the boundary layer remain under a critical value as
long as possible. The transition of the laminar flow in the boundary layer to
the turbulent form will than take place as late as possible and may possibly
even occur without the formation of a disturbing laminar bubble. This is
nowadays often the case because of the specific shape of the modern profiles,
where the thickest part may even lay beyond the middle of the total depth of the
profile. At the upper side of the wing transition takes place at about 65%
+/-10% of the profile depth, depending on flap position and velocity. At the
lower side of the wing, the profile is nowadays almost flat with only some
detailing in the shape at the rear end to promote a smooth run-off of the air
flow. As a result the boundary layer remains laminar up to about 85% +/-10% of
the profile depth. Than, by using ZZ tape or smalll blow holes in the profile
surface the flow is artificially made turbulent to avoid uncontrolled laminar
detachments of the flow with undesired suction pockets.
- Effect of the boundary layer on the pressure
distribution
The thickness of the viscous boundary layer for a
flat plate can be calculated fairly accurate with quite simple formulas for
both laminar and turbulent flow. The result gives a useful indication of this
thickness for wing profiles. At the nose of a profile the boundary layer is
still quite thin. At 0,5 m profile depth this thickness is about 2, 2 mm for
the average velocity of the glider with laminar flow. With turbulent flow
existing already at the trailing edge, e.g. because of bugs or water droplets,
than the thickness of the turbulent boundary layer at 0,5 m downstream may
already have a thickness of about 11 mm and that is five times as much as that
for a laminar boundary layer. So a turbulent boundary layer increases
downstream on the wing profile much more in thickness than a laminar boundary
layer. The boundary layer displaces the streamlines adjacent to the profile
over a certain distance away from the surface. By adding this so-called
displacement thickness to the contours of the profile itself, the aerodynamic
thickness of a profile is in practice therefore slightly larger than the
geometric thickness. This effect on the geometry is of course taken into proper
account when calculating velocity and
pressure distributions just outside the boundary layer and is in fact the
reason why a profile experiences pressure drag in addition to frictional drag.
This explains the earlier in this article mentioned paradox of d'Alembert. The
effect of the boundary layer is clearly reflected in the pressure distributions
shown in figure 9. At the continuous curve with a viscous boundary layer the
transition from laminar to turbulent is well recognized; the dotted curve with
only inviscid flow has no boundary layer and does therefore not show this
phenomenon.
Figure 9: Effect of the boundary layer on the pressure distribution
- Friction drag and pressure effects
The friction coëfficient on the surface of the
profile can be calculated quite well with well known boundary layer velocity
equations for both laminar and turbulent flow. These results show show that for
the usual velocity range of gliders the friction effect of turbulent flow is
typically a factor 6.5 higher than that for laminar flow. It is therefore
recommended, using a good design of the shape of the wing profile, to keep the
boundary layer as long as possible laminar both at the top and the bottom. It will
also be clear that especially near the nose of the wing, with its typical flow
stagnation effects, the wing surface must be absolutely clean knowing that
distortions only as high as 0,1 á 0, 2 mm will change the flow downstream in
the boundary layer from laminar to turbulent earlier than intended. Because of
spreading out effects, larger partions of the wing surface will suffer from
high friction drag.
Although a turbulent boundary layer sticks better
to the profile surface than a laminar one, the former can leave the surface
also quite easily if the pressure downstream rises too quickly. This happens
quite naturally at the trailing edge of the profile when flying too slow for
the selected flap position. Than extra pressure drag will result from turbulences
and suction effects behind the profile. The profile drag coëfficient CDp increase than rapidly as is shown in figure
10 for the higher values of the lift coefficient CL.
Incidentally, the drag of the profile increases also quickly when flying with a
too small angle of attack, so too fast at the selected flap position. In this
case, the transition from laminar to turbulent flow in the boundary layer
starts directly at the front lower side of the profile. This is clear from
figure 10 for the lower values of CL .
It will be
clear that the pilot must always take care of operating his glider in the low
drag bucket such that the transition of the flow in the boundary layer only
happens as far backwards as possible,
both at the top and at the bottom of the profile.
Figure 10: Profile drag for a modern profile with flaps (REF. LB,
TUDelft)
Listening
to the boundary layer
- Laminar versus turbulent
It is very important to fly with a laminar
boundary layer as long as possible both at the top and the bottom of the wing.
With an existing profile pilots have this largely under control themselves by
flying with a very clean and pure wing surface, but above all, by a correct
combination of flap setting and speed. At laminar flows air particles in the boundary
layer move nicely parallel to each other. They "see" each other but
do not bother each other in their movements, so there is little energy transfer
among them and to the wing surface. With turbulent flow those movements are
pretty irratic and there is quite some exchange of kinetic energy between the
air particles among themselves and with the wing surface. If you listen with a
small microphone to the flowing air in the boundary layer, this as a follow up
of previous measurements with the ASW-19 of TUDelft, then you can very well
perceive if the flow is laminar or turbulent at the measuring spot. In 2008 I
did that for our Ventus XT at various combinations of velocitys and flap
settings, both during straight flight and circling. The method works excellent
and fulfills the objective of keeping the flow across the wing surface laminar
as long as possible, both at he upper and lower side of the profile. For the
measurements I used a very small microphone of the type that is used in hearing
devices and made available to me by Sonion. A very thin measuring tube
(flexible tube) with a length of about 100 cm picks up the signal and leads it
to the microphone. In this way, the flow is not disturbed by the set up. A
small microphone amplifier is mounted in the fuseslage of the glider. It works
on the 12V board voltage and is connected to a small speaker box in the luggage
compartment. The box is well audible to the pilots ear in front of it.
- Measurements on the upper side of the profile
Figure 11 : Measuring tube on top of the wing
When listening at the upper side of the profile,
the measuring tube was mounted about 70 cm aside from the fuselage. The tip of
the measuring tube was positioned at 60% profile depth (48 cm from the leading
edge). Figure 11 shows this setup.
I was expecting that at this depth and a correct
combination of velocity and flap setting the flow in the boundary layer would
just be laminar and this was indeed the case. Fguur 12 gives a presentation of
the observations. Above the curve the boundary layer flow at the tip of the
sensor is turbulent (e.g. at 110km/h and flap position "– 1"). Bbelow
the curve this flow is nicely laminar (e.g. at 110/h and flap position "+
1"). On the curve itself transition takes place (e.g. at 110kmh and flap
position "0")
Figure 12 : Sensor on top of the profile; observations for various
combinations of flap position and velocity
Generally speaking one can say that at a correct
combination of flap position and velocity (e.g. 0 at 110 km/h), the noise
picked up at the tip of the measuring tube and fed to the small microphone is
typically laminar in character. So hardly noticeble. When selecting a more
positive flap position (eg. + 1) the transition moves in a convenient way more
backward on the profile because at the larger flap deflection the pressure
further downstream decreases a little more. When selecting a less positive flap
position (e.g. from 0 to-1 to 110 km/h) the sound perception with the
microphone is typically turbulent. Than the transition has moved in a less
favourable sense more forward on the profile.
The observations for the position of the
transition on the upper surface of the profile as a function of the flap
position and velocity correspond well with results of detailed calculations
performed at TUDelft and Uni-Stuttgart.
- Measurements at the bottom of the profile
Similar measurements as described above were
performed at the lower side of the profile at several positions in front of and
after the ZZ tape. A detailed discussion of this interesting measurements will
mot be given here.
Togeteher with the measurements on the upperside
of the profile I could quite accurately determine at which combinations of flap
position and velocity an as long as possible laminar boundary layer occurred .
The first and second line of Figure 13
(flap position and velocity Vv) give the results of those favourable
combinations. The velocities listed in the three lines there below were calculated
from these results and apply to higher wing loadings and larger bank angles.
Compared with the advice of the manufacturer of the glider, there is good
agreement. By listening to the noise of the microphone at different wing
locations I came automatically to favourable combinations of flap position and
flight velocity. With transitions too far upstream, i.e. before the 60% wing
cord location at the upper side of the profile respectively before the 85%
position at the lower side of the profile, velocity and/or flap position had
better be adjusted. The flaps must always be set for a given velocity in such a
way that the boundary layer both at the upper side and under side of the
profile is laminar as long as possible thereby obtaining the low drag
co-efficiënt of about 0,005 shown in Figure 10.
Figure 13: Convenient combinations of flap positions and
velocities determined from listening to the boundary layer
The microphone was also used to listen to the flow
in the boundary layer at the underside of the profile just downstream of the ZZ
tape at about 90% profile depth. This gave a good impression of the operation
of the ZZ tape at the different combinations of velocity and flap positions. A
clear audible difference in the character of the noise showed between the
spontaneous "unstable" turbulence that had emerged in front of the
position of the ZZ tape and the "stable" turbulence created by the ZZ
tape when the still laminar flow was forced to turbulent.
-Flying through turbulent air
Large-scale turbulences in the air cause
spontaneous changes in the angle of attack of the flow hitting the profile and
reactions to that of the character of the flow in the boundary layer. The
abrupt changes in the location of the transition on the profile is very good
detectable with the sensor. The same effects are noticed when small but pretty
abrupt rotational movements around the dwarsas of the wing are created using
the control stick. Both events cause extra profile drag and are therefore
undesirable. Also from Figure 10, it is clear that abrupt increases or
decreases in the lift coefficient, which are a consequence of these changes,
can cause significant increases in profile drag.
-Permanent use of the turbulence sensor
Also this season the "turbulence detecting
system" was used in our XT, but now with the sensor only on the upperside
of the wing at 55% profile depth. When circling in turbulent thermals, I like
to make sure that the current working point in the lift curve is still at a
convenient place. When a too frequent turbulent flow occurs at the sensor
position, than the transition happens too far forward on the profile and on
average I fly with a too large angle of attack, so too slow or with not enough
positive flaps. Than the Cl-Alpha working point is too close to or maybe
already in the so-called "step" of the Cl-Alpha curve of wing
profiles with flaps. The negative effect on descent rate is than high because
of turbulences in the air. An article abouth this effect was published earlier
(http://home.planet.nl/~kpt9/The_Step.htm).
Remedy is to fly a lttle faster and with more banking angle and flaps. However
this is on the expense of additional polar sink rate, so only do it if really
needed. A subtle interpretation and evaluation of effects is conclusive here.
The option of circling very slowly in turbulent thermals, i.e. with the
workpoint past the “step”is something, I would like to investigate further. In
that earlier article I indicated that this seems promising (see Figure 6 in the
publication). The signal of the sensor will constantly indicate than a
turbulent condition at the tip of the sensor. The future will tell whether this
slow flying method works while having a large profile drag and if the method
can be performed safely in turbulent thermals.
-Suction of the boundary layer
Wind tunnel research, e.g. at TUDelft, has shown
that boundary layer suction is a promising method:
- to keep a laminar boundary layer laminar and
attached to the profile over the entire profile depth
- to keep a turbulent boundary layer attached to
the profile up to the trailing edge.
Both these effects reduce the drag of the profile
significantly while the lift co-efficiënt increases. As a result, gliders with
boundary layer suction perform more than 30% better, an improvement which is
certainly not possible when optimising usual profiles because of physical
limitations. This promising possibility suffers of course from technical
problems, particularly the unavailability of suitable porous plate material.
Also aerodynamic problems need a lot of attention. The energy necessary for a
suction pump to to keep a laminar boundary layer indeed laminar and bound to
the surface is relatively low. Only the very lowest part of the boundary layer
needs be sucked off to keep the flow rate also close to the profile greater
than zero. This will prevent the boundary layer from detaching from the surface
and become turbulent as described earlier in this article. With effective solar
panels on the front part of the wing, the required suction energy can be
generated.
To keep a turbulent boundary layer attached to the
profile surface more air must be extracted and therefore much more pump energy
is needed. This can be problematic in gliders. But this is only required at
larger angles of attack during thermalling where the boundary layer becomes
already turbulent in front of the extraction area. It seems possible with a
slightly adjusted profile to just prevent this from happening so that still
only low power is required to keep the boundary layer laminar and attached to
the profile up to the trailing edge. Loek Boermans (LB) of TUDelft has the
objective to make boundary layer suction to a success. Currently he works hard
on his projext and makes good progress. In Thermiek 4 – 10 he reported already
quite extensively on this.
-Finally
"Measuring is knowing" is the logo of my
former employer; I still think in that way. However a good theoretical
knowledge of what you observe is obviously necessary to get to correct
conclusions. In a separate document I have tried to support this statement. The
study of a significant amount of literature in the field of aerodynamics and my
good contacts with the well-known aerodynamicist Loek Boermans (LB) of TUDelft,
which assisted me quite extensively
with this article, has certainly helped me to understand what I am writing I
hope.
- Who are we
Boermans : (Prof.) Ir.
Loek Boermans is a Dutch Aerodynamicist and associate professor at TUDelft. He
has acquired worldwide fame with the designs of very good performing low
velocity wing profiles, especially those for gliders. For many years he is now
President of OSTIV. Although retired he spends currently much attention to boundary layer suction to
keep this thin layer of air as long as possible laminar and prevent detachment
with the aim of keeping the profile drag as low as possible. Loek was born in
1946 at Venlo.
Termaat : Ir. Karel Termaat was a
reactor physicist at KEMA Arnhem. He was one of the pioneers at the now closed
Dodewaard Nuclear Power Plant. After his retirement he took further interest in
the aerodynamics of gliders. In Loek Boermans he found a dedicated teacher. As
an experienced cross country glider pilot and instructor he tries now to form a
link between the theoretical work of Loek Boermans at Delft University of
Technology and the aerodynamic imaging of glider pilots. Karel was born in 1936
at Woerden.
Delft /Arnhem
April 2012
Karel Termaat
